Forum Discussion

dragonflymr's avatar
dragonflymr
Icon for Cirrostratus rankCirrostratus
May 23, 2017

Web Acceleration profile and changing HTTP/1.1 to 1.0

Hi,

I am not HTTP expert but this behavior is quite a surprise to me - but maybe it's perfectly OK?

Setup (tested on v11.2.0HF7):

VS with Web Acceleration profile attached (based on optimized-caching), no OneConnect, no HTTP Compression Profile

Request from client:

GET /images/wwfr.png HTTP/1.1
Host: cache.test.com
User-Agent: Mozilla/5.0 (Windows NT 6.1; WOW64; rv:39.0) Gecko/20100101 Firefox/39.0
Accept: text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8
Accept-Language: pl,en-US;q=0.7,en;q=0.3
Accept-Encoding: gzip, deflate
Connection: keep-alive

Request from BIG-IP to backend server

GET /images/wwfr.png HTTP/1.0
Host: cache.test.com User-Agent: Mozilla/5.0 (Windows NT 6.1; WOW64; rv:39.0) Gecko/20100101 Firefox/39.0
Accept: text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8
Accept-Language: pl,en-US;q=0.7,en;q=0.3
Accept-Encoding: gzip, deflate
Connection: keep-alive

If I will disable Web Acceleration profile then request from BIG-IP to server will be HTTP/1.1.

So why such change?

Another issue - I have iRule attached to VS to disable caching response on the client side:

when HTTP_RESPONSE {
    HTTP::header insert Cache-control no-store
} 

What surprised me is that BIG-IP is taking this header into account when deciding if server response is cacheable. That is quite strange as header is inserted on client side so should not influence server side decision - or I am wrong here?

Piotr

  • OK, should check AskF5 first. Still this article The BIG-IP WebAccelerator virtual server requests are translated to HTTP/1.0 is not explaining why caching profile needs to change requests to HTTP/1.0?

     

    Why caching content on BIG-IP requires HTTP/1.0 request to server?

     

    According to articles I found server is not obliqued to send HTTP/1.0 reply when receiving HTTP/1.0 request - mine is replying with HTTP/1.1 (Apache2). Only limitation seems to be that server should not use incompatible headers when replying using HTTP/1.1 - but not sure if it's a must or not.

     

    Piotr